Transcript The Fourth Computational Intelligence Reading of IEEE SMC

An Evolutionary Approach to
Multiobjective Clustering
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. 11, NO. 1, 2007
Julia Handl and Joshua Knowles
Speaker: 陳進和
2007年8月1日
Outline




Introduction
MOCK (Multiobjective clustering with automatic k-determination)
Experimantal results
Conclusion
1. Introduction

Assess the performance of clustering
algorithm



Lack of a formal definition of clustering
No objective performance criterion
Multiobjective optimization is used to tackle


Unsupervised learning problem
Data clustering
2. MOCK
(Multiobjective clustering with automatic
k-determination)

MOCK consists of two phases


Phase 1: Initial clustering phase
Phase 2: Model-selection phase
Phase 1: Initial clustering phase

PESA-II



Internal population to explore new solutions
External population to exploit good solutions
Objective functions
Phase 1: Initial clustering phase

(cont.)
Genetic representation and operators



Locus-based adjacency representation
 No need to fix the number of clusters
 Well-suited for standard crossover operators
Uniform crossover
 One-point or two point
Neighborhood-biased mutation operator
 Quickly discard unfavorable links
 Explore feasible solutions
Phase 2: Model selection

Motivating concepts

Inspired by Tibshirani et al.’s Gap statistic, a
statistical method to determine the number of
clusters in a data set
3. Experimantal results

Parameter setting
4. Conclusion

MOCK


Outperform traditional single-objective clustering
techniques
Keeping the number of clusters dynamically